The rise of powerful AI will be either the best, or the worst thing, ever to happen to humanity. We do not yet know which."
—Stephen Hawking
Technical breakdowns and news on newly released AI models, multimodal capabilities, benchmark evaluations, and parameter scaling.
The appointments of Michael Rogers and Tom Moran add senior US military and intelligence expertise to the Japanese AI company’s efforts to address emerging security challenges associated with artificial intelligence (AI) and its applications in national security.
Block has introduced Dessa, a speech-to-speech AI model trained from scratch on human conversation without text data or pretrained models. The company says the model can listen and speak simultaneously, interpret vocal cues and respond more naturally during real-time conversations.
OpenAI has acquired four-year-old Orchestra Bio, a startup building AI agents to help pharmaceutical companies manage drug development. The company plans to bring Orchestra Bio’s team and technology together with Rosalind, its AI models and tools for life sciences research. “Our goal is to help scientists make discoveries and manage the work needed to turn… Continue reading OpenAI Acquires Orchestra Bio to Advance AI Drug Development with GPT-Rosalind
TypeSafe AI’s decision-focused model Jev reached 35.3% of paid teams on Vercel AI Gateway within its first 14 days, setting a record for adoption. The model also recorded a 55.6% week-two return rate, while its share of inference spending remained small, according to Vercel’s October 2026 AI Gateway Production Index.
Former Google DeepMind researchers’ Reflection AI is positioning Beam against leading Chinese open-weight models on coding and agentic capabilities, while highlighting its lower inference compute and efficient use of model parameters.
TypeSafe AI has reopened sign-ups for Jev, its “System One” AI model, a week after suspending registrations following a surge in demand. The company has temporarily withdrawn the $5 free credit for new users, citing misuse by “a few bad actors.”
Google DeepMind and Google Research have expanded their WeatherNext family of AI weather models with WeatherNext Cyclones, a system designed specifically for tropical cyclone forecasting. The model predicts a cyclone’s track, intensity and wind structure. Google says its three-day forecasts are as accurate as what prior models could provide for the next two days, giving… Continue reading Google Advances WeatherNext AI From General Weather Forecasting to Specialized Cyclone Predictions
Teen searched OpenAI’s ChatGPT for scenarios about harming his family before his mother and brother were found dead, echoing ChatGPT’s alleged role in the 2025 Florida State University (FSU) mass shooting FSU mass shooting and raising fresh concerns about generative AI’s risks to vulnerable users.
OpenAI said its upcoming AI model Astra has shown “critical” cyber capabilities under the company’s internal framework and will take longer to develop with necessary safeguards.
Tech giant Meta has confirmed that one of its AI models accessed the internet and hacked into a third-party company’s systems during a cybersecurity test after a misconfiguration error by independent testing partner Irregular. The incident places Meta alongside OpenAI and Anthropic, whose models also recently breached external organizations while under evaluation.
